my biggest trade

Sorry for not having posted more regularly but I have been busy setting up my biggest trade of the last few years….the sale of my house.  My family and I are indeed relocating to sunny South Florida over the summer and I have been actively preparing to put our house on the market.  This is an even greater challenge given that we are now in one of the worst housing recessions of the last 20+ years.  House prices are falling, credit is scarce, and house equity is melting away.  Our only hope is that while we may lose a few feathers on our OH house, we will get a sweet deal on a FL house, all in all this may still turn out to be a favorable move.

The big decision that we had to make was to either hire a real estate agent or sell it ourselves (FSBO).  About three years ago we were in the same situation with our house in PA and decided to give FSBO a chance.  Of course the market was completely different at the time but we struck gold and sold our house in 24 hours for the full asking price, ending up only giving away a meger 1.5% commission to the buyer’s agent.  We are not expecting the same kind of success story in this case but we feel that the 6%-7% that we would be charged as sellers would only add insult to the injury of falling house prices, and justifies once again the FSBO approach.  This time I listed my house on www.fizber.com and purchased their MLS ad package.  With this package my house will be listed not only on Fizber.com, but also on 80 of their partner sites, and last but not least in the MLS database and www.realtor.com.  I am hoping that this added visibility will attract the right buyers.  Wish me luck!

This being said, the addition of a full time job and the time it takes to put a house on the market while taking care of my family has left little to no time for trading.  I haven’t even logged into Tradestation this week.  My only contact with the market has boiled down to listening to Bloomberg on my way to work. 

self-adaptive Bollinger Bands

One always hesitates in deciding what value should be used for the length of the mean and for the standard deviation (STD) of Bollinger Bands.  Very often we will see generic values such as 20 and 2.0 being used, respectively.  The idea behind self-adaptive Bollinger Bands is to stay away from static STD values and adapt them to the market movements.  Our goal in doing so is to ensure that price movements remain contained, in 95% of the cases, within the upper and lower Bollinger Bands.  This 95% containment being the purpose of the conventional use of 2 STD from the mean in the first place. 

The self-adaptive indicator that I am posting here reassesses the optimal STD value every 100 bars.  It simply verifies if over the last 100 bars, 95% of the price closes have remained within the bands.  If less than 95% of close were within the bands then the STD is increased by .1 (bands are expanded), otherwise it is decreased by .1 (bands are contracted). 

The example below show the adjustments that were made to the standard deviation over a sample of approximately 1,300 bars on @ER2, starting with a STD value of 2.0.  The example shows that in this sample the value of 2.0 was insufficient in reaching the target of 95% of the population, therefore the standard deviation was adjusted upwards to about 2.5 to reach the 95% containment target.  In the list below the first number (e.g. 0.84) shows the percentage of the last 100 bars that closed between the Bollinger bands, while the second number (e.g. STD 2.10) shows the adjustment that was made to the standard deviation as a consequence.  We can see that eventually this sample stabilizes around an optimal standard deviation value of 2.5.

0.84 --> STD 2.10
0.87 --> STD 2.20
0.85 --> STD 2.30
0.90 --> STD 2.40
0.98 --> STD 2.30
0.88 --> STD 2.40
0.88 --> STD 2.50
0.93 --> STD 2.60
0.95 --> STD 2.50
0.93 --> STD 2.60
0.93 --> STD 2.70
0.96 --> STD 2.60
0.96 --> STD 2.50

The code for this self-adaptive version of the Bollinger Bands is provided below:

This code could be further improved by allowing a more rapid increase of the STD values when we observe a significant delta from the target containment percentage.  Indeed as we see in the example above, it took approximately 500 bars to eventually reach an STD setting that offered a containment >= 95%.  Accelerating the rate of increase in the STD could significantly shorten this catch-up period.  Note that this could also be improved by shortening the reassessment period from 100 bars to 50 bars for instance. 

Another potential improvement effort could also be focused on the number of bars used in the calculation of the simple moving average.  In this case we have used a conventional 20-period SMA as the basis for the Bollinger bands but here also an adaptive system could be developed to optimize the length, for instance as a function of market volatility.  As market volatility increases (e.g. as indicated by an increase in the standard deviation from one bar to the next as measured over the last 30 periods), we would lengthen the SMA period.  On the other as volatility contracts, we would shorten the SMA period.

The appealing aspect of self-adaptive systems is that they respond to current market movements and are therefore by default more ‘forward-looking’ that parameter optimizations based on historical data.

new twist on floor trader pivots

Inspired by the article “Fibonacci Pivot Points” published in the February 2008 edition of Futures & Options Trader (FOT), I have developed a Tradestation indicator that is capable of displaying both the standard floor trader pivots and the Fibonacci pivot points.  Calculation of the central pivot is the same for both techniques.  The differences appear in the calculation of the different support and resistance levels.

Standard Floor Trader Pivots:

Central Pivot (PP) = (H + L + C)/3
First resistance level (R1) = (PP*2) – L
Second resistance level (R2) = PP + (H – L) 
First support level (S1) = (PP*2) – H
Second support level (S2) = PP – (H + L)

Fibonacci Pivots:

R3 = pivot + ((H – L) * 0.764)
R2 = pivot + ((H – L) * 0.618)
R1 = pivot + ((H – L) * 0.382)
Pivot = (H + L + C) / 3
S1 = pivot – ((H – L) * 0.382)
S2 = pivot – ((H – L) * 0.618)
S3 = pivot - ((H – L) * 0.764)

Where H, L, and C are yesterday’s high, yesterday’s low, and yesterday’s close, respectively.

In my Tradestation implementation you can simply switch between standard floor pivots and Fibonacci pivots by switching the PivotType value from a 1 to a 2 in the input settings.

TS Code: TG_Pivot.ELD

The article published in FOT indicates that Fibonacci pivot points tend to be better predictors of support and resistance levels than the conventional floor trader pivot technique.  This was proven true a strategy backtesting based on countertrend setups.  This backtesting showed that, especially in volatile markets, fib-based pivot points offered better profit targets and therefore higher potential profits.

This being said these pivot levels are simply S&R level predictors and their popularity seems to be decreasing in the trader community, which in turn could mean that they are less likely to work as less people pay attention to them.  Personally I still like to include them on my charts and use them as potential profit and stop loss targets.  But in my view they are not necessarily as meaningful as simple high, low, and close levels from the previous day or week or even the spontaneous S&R levels that form along obvious price congestion areas during the intraday session.